Forecasting the rheological properties of alkali-activated concrete utilizing gene expression programming

Accurate prediction of fresh-state properties is essential for understanding the rheological behaviour of alkali-activated concrete (AAC), as these parameters directly influence mix uniformity and overall material performance. Gene expression programming (GEP) effectively captures complex relationships between mixture components and material properties, offering superior flexibility, interpretability, and adaptability compared to traditional regression and black-box models. It generates explicit mathematical equations suitable for engineering design while automatically selecting relevant features. Accordingly, this study employed gene expression programming (GEP) to develop a robust predictive model for the slump (SL), dynamic yield stress (DYS), static yield stress (SYS), and plastic viscosity (PV) of AAC, eliminating the need for time-consuming and resource-intensive physical experiments. The GEP model demonstrated excellent performance, achieving a coefficient of determination (R²) of 0.979 and a root mean square error (RMSE) of 1.183 for SL, and an R² of 0.960 with an RMSE of 3.293 for SYS. Similarly, strong predictive accuracy was observed for DYS (R² = 0.962, RMSE = 0.161) and PV (R² = 0.962, RMSE = 5.266). The study also derived explicit equations using GEP for practical use. To enhance accessibility, a user-friendly graphical interface was developed, enabling instant predictions from input parameters. The GUI simplifies mix design, minimizes testing requirements, and speeds up decision-making, thereby promoting the adoption of AAC in construction and ensuring improved control over fresh-state properties.